Using LiDAR-Derived Information on the Vegetation Canopy Structure to Scale ET Estimates Beyond the Tower Footprint in the Western Boreal Plains, Alberta, Canada

In Canada a large portion of the boreal forest is comprised of the Western Boreal Plains (WBP)—a region characterized by a mosaic of vegetation cover and persistent water deficit conditions experienced on a 10-15 year cycle. Owing to this, the WBP is a hydrologically sensitive region and future changes to the climate will significantly impact the region's water balance. The WBP is also an economic hub for the country's natural resource extraction, creating significant disturbances on a hydrologically sensitive landscape. Sparse environmental measurement stations also make it challenging to gain information on water balances within different land cover types used to drive and validate models outside of the tower-footprint scale. Remote sensing can provide spatial estimates of land cover type that can be used to better understand vegetation characteristics which influence the water balance of the WBP—though further investigation is needed to determine the most appropriate resolution at which remote sensing data is best collected. Additionally, large disparities in scale between point or tower validation data and coarse satellite-derived evapotranspiration (ET) estimates can introduce error into validation processes.;This study uses high resolution (1m x 1m) Light Detection and Ranging (LiDAR)derived vegetation parameters and measured hydro-meteorological variables to estimate spatially variable ET at a 5km x 5km study site in the WBP. High resolution ET estimates are scaled to lower resolutions to provide insight into the accuracy of ET estimated from low resolution satellite systems, as well as to bridge the gap between fine-scale validation data and coarse ET estimates characteristic of hydrologic components of GCMs. Environmental drivers and estimates of ET were scaled up (5km x 5km) from the point scale using spatially variable and spatially averaged methods. ET estimates scaled using spatially variable vegetation parameters overestimated measured values by 1.5mm, or 3% (r2 =0.51, RMSE = 0.14mm d-1). ET estimated scaled using averaged vegetation parameters overestimated measured values by 3.3mm, or 6% (r2 = 0.49, RMSE = 0.37mm d-1 ). Subsequent scaling of 1m x 1m spatially variable ET estimates to lower resolutions characteristic of global satellite systems—i.e. SPOT (10m), Landsat (30m), and MODIS (250m, 500m)—yielded increasingly inaccurate results ranging from overestimations of 3.2mm (7%) at 4m resolution, 3.4mm (8%) at 10m, 3mm (6%) at 25m, to underestimations of 3.5mm (7%) at 250m, and 4.5mm (9%) at 500m when compared to measured values.